[02:54:35] legoktm: what about making an experimental on-demand 7.2 job, then once it passes replacing 7.1 with it? [02:55:21] MaxSem: 7.2 is in experimental queue as of an hour ago :) [02:55:49] WHAT ELSE DID YOU READ FROM MY BRAIN WITHOUT PERMISSION? [02:56:32] hehe [02:58:00] I have a script that'll trigger tests against all repos that I can run over night [02:58:33] oops, https://integration.wikimedia.org/ci/job/wmf-quibble-core-vendor-mysql-php72-docker/1/console has an interesting sidebar [03:02:44] that's new [03:03:02] I think Krinkle was the one setting those up? [03:03:38] Also, why are there suspiciously few failures? [03:04:05] hehe, must be a regex fail of some kind [03:04:29] they've been around since 2013 the Console collapsable sections, but I hope the regexes from time to time to match the current output. [03:06:01] MaxSem: it's only running the databaseless tests [03:06:25] it runs the @group Database tests in a later stage [03:10:08] MaxSem: fixed the npm-ls regex [03:10:35] Whee [03:11:01] the patterns are hard to get right because the "safe" lines to match often have color codes on them which breaks the whole output formatting if matched. [03:11:21] So I usually have to match the first line of output instead, which isn't always safe. [03:11:36] In this case, I made it assume that anyline starting with /srv/workspace is the npm-ls tree [03:11:50] Anyway, it's a mess. [04:08:12] legoktm: hm.. phan not found for echo? [04:08:36] looking into it right now [04:08:48] https://phabricator.wikimedia.org/T206893 [04:11:08] legoktm: cool. So looks like the fail is good. What changed :) ? [04:11:30] 9a9cb346a8b5b2e6ca122a5d49b222c95fa9f722 [04:12:14] Ugh, thats not an easy one to fix. [04:13:14] Is the job easy to disable? [04:13:40] I think we can move the check into the dockerfile [04:25:08] legoktm: btw, the gadgets 2.0 thing, there's apparently some VPS instances for that still. Totally fine, but just checking if you use them still and/or if it's hard to re-create later in context of T204692. If I recall correctly, that was for public demos, not for debugging, given we now have dev tool for multiple wiki locally, but either way is fine, upto you :) [04:25:08] T204692: cloudvps: social-tools project trusty deprecation - https://phabricator.wikimedia.org/T204692 [04:51:20] fixed the phan issue [04:51:25] Krinkle: I think it was just demos... [06:32:50] soooo with https://gerrit.wikimedia.org/r/465201 PHP 7.2 on core is passing [06:46:03] aaaaand a clean run on travis-ci!! [07:10:35] Makes you wonder what about all those failures previously reported for 7.2 [07:11:44] I closed all of them that were just on travis except one of them that Reedy reproduced locally [07:13:41] The one I couldn't repro? [07:13:46] yeah [10:49:14] It's possible it's older PHP point release bugs too [10:49:18] Stuff they've fixed under the hood [10:50:18] migrateActors is trying to insert duplicate rows... [10:50:28] Anyone against just making it do INSERT IGNORE? https://gerrit.wikimedia.org/r/467073